工艺流程技术的重要性

      从我之前的两篇关于摩尔定律的博文中，我想这篇文章的题目在某种程度上已经不言自明。在前面的文章中我谈到了物理学改变了我们进行IC设计的方法。至少对我而言很明显的一点就是：作为工艺流程技术的领导者，理解并满足新型数字（更不用说模拟了）设备对SoCs的需求是相当重要的，，达芬奇技术就是个典型的例子。

      晶圆厂的所有权是关键。这是个很热门的话题：许多公司把他们的制造工作外包给中国大陆或台湾地区。我可以马上说：决定做工艺流程技术很容易，但是真正做好就难了。在选择技术合作伙伴时，这家公司生产的芯片能否在他们自己公司的实验室实现非常重要。

      在没有商业目的的前提下，TI已经成为工艺技术的领导者，自我在TI工作的时候，TI就有了自主的晶圆厂。现在主要用130nm工艺生产，90nm已经很好地在进行大规模生产，65nm则在一些新设备上应用较多。45nm即将启动，距离成功还有很长的路要走。

      问题是我们现在把数以百万计的晶体管安装在硅片上，如果我们不能确保所有部件可以以最优化的方式协同工作，那我们的工作就白费了。一家外包的晶圆厂不可能与那些我之前提到的投入很大的企业在知识水平上和对物理的理解上可以相提并论。更重要的是，在IC设计和工艺流程设计之间还存在着一个纽带。你不能因为一个而抛弃另一个：正像爱情和婚姻，马和马车一样密不可分。

      所以，继续工艺流程设计创新对我们非常重要，但这些工作并不能凭空完成。因为异花授粉的工作只有当一个公司同时懂得应用设计师面临的问题，IC设计师面临的问题，以及晶圆厂涉及的问题时才可能做得到。哪怕是一个极小的疏忽就可能造成数百万美元的损失，数千名员工以及几百位管理者的头发都会掉光(当然秃头的经理除外),也就是他们的努力就可能会付之东流。

      所以我一直坚信，选择一个既懂得工艺流程技术，又了解芯片设计的供应商，是我们系统设计师成功的关键。

                 The Importance of Understanding Process Technology

From my last two blogs regarding Moore’s Law, I think the title of this entry is somewhat self evident. In those postings, I talked about the physics that are changing the way we approach IC design. The obvious (to me anyway) offshoot of that is that being a leader in process technology is vital to understand and meet the demands of the new digital (not to mention analog) devices that thrive on SoCs: DaVinci technology is an obvious example.

Ownership of the fab is the key. This is a hot-button issue given that many companies outsource fabrication to countries like Taiwan and China. Let me say at once that it is easy to decide to do process technology, it’s hard to do it well. Whether a company can actually produce the chips that work in its own lab environment should be an important factor when choosing a technology partner.

Not to turn this into a commercial, but TI has been a leader in process technology and has owned fabs as long as I have worked there. Currently, 130nm is the workhorse in production, 90nm is well into mass production, and 65nm is ramping up for a few devices right now as well. The 45nm node is on the horizon, and we still have a long way to go before we run out of room.

The bottom line is that we are putting millions of transistors on the silicon, which would be worthless if we couldn’t be sure that all the components work together optimally. An outsourced fab cannot possibly have the same level of knowledge as a company that has invested heavily in understanding the physics I have outlined previously. Furthermore, there has to be a tie between IC design and process design. You really can’t have one without the other: love and marriage, horse and carriage and all that (My apologies to Frank Sinatra).

So, continuing to drive process design innovation is a key for us. But it cannot be done in a vacuum. There has to be the cross-pollination that can only occur within a company that understands the issues facing application designers, the issues facing IC designers and issues regarding the fab. Even the smallest disconnect can mean the loss of millions of dollars, thousands of staff hours and hundreds of hairs from the heads of management (bald managers excluded).

So, I am a believer that choosing a vendor that understands process technology as well as chip design will be a crucial part of my success as a system designer.